Opened 5 years ago

Closed 5 years ago

Last modified 5 years ago

#13103 closed bug (duplicate)

gawk triggers a segfault in ICU while building PCI header file

Reported by: anevilyak Owned by: korli
Priority: normal Milestone: Unscheduled
Component: Kits/Locale Kit Version: R1/Development
Keywords: Cc:
Blocked By: #12515 Blocking:
Platform: All


This is x86_gcc5 hrev50738. When attempting to build a jam -q @image, gawk consistently fails while attempting to build the PCI header file from pci.ids. After tracking this down a bit more, the crash itself seems to be triggered somewhere in libicu as indicated in the attached crash report. I'm uncertain right now if this is a problem in the gawk port, or ICU or something else. I'm also currently uncertain as to what revision this started with, as I don't generally do full image builds from the OS itself (this was found while testing something unrelated for Axel).

Assigning this to locale kit for now, until a more suitable culprit can be found.

Attachments (1) (10.5 KB ) - added by anevilyak 5 years ago.
gawk report when stopped at the received SIGSEGV.

Download all attachments as: .zip

Change History (6)

by anevilyak, 5 years ago

gawk report when stopped at the received SIGSEGV.

comment:1 by korli, 5 years ago

Sounds like #12515.

comment:2 by anevilyak, 5 years ago

Blocked By: 12515 added
Resolution: duplicate
Status: newclosed

Indeed, sorry I missed that one. Odd that it's compiler-specific though.

comment:3 by korli, 5 years ago

you think it works on x86_gcc2? It fails on x86_64 too.

comment:4 by anevilyak, 5 years ago

I had jua try it on gcc2 and it worked there. This is also mentioned in #12515 ( ), which seems to corroborate that jua's case wasn't an isolated coincidence.

comment:5 by anevilyak, 5 years ago

I should also note, updating to gawk 4.1.4 from HaikuPorts made no difference.

Note: See TracTickets for help on using tickets.